PsycNET®

Prevention & Treatment. 1998 Jun Vol 1(2)

Posted 06/26/1998. In addition to responding directly to the issues raised by the commentators, I compare the Kirsch and Sapirstein (see record 1999-11094-001) data to those of 2 other meta-analyses (Joffe et al., 1996; Walach & Maidhof, in press) of antidepressant medication. The 3 meta-analyses used different inclusion criteria and different methods of calculating effect sizes. One is confined to long-term effects of treatment and another to acute effects. One is based on number of patients clinically improved and the others on standardized change scores. In addition, one included data allowing for calculation of response rates based on intent-to-treat criteria. Despite differences in study characteristics and methods of analysis, all revealed that the a large proportion of the response to medication was duplicated by placebo, and all revealed exceptionally high correlations between the placebo response and the drug response. These data reveal that the findings we reported are reliable and generalizable, and they indicate a pressing need for new methodologies in clinical trials. [In reply to L. E. Beutler, L. P. Rehm, R. M. Dawes, and D. F. Klein, see records 1999-11094-002, 1999-11094-003, 1999-11094-004, and 1999-11094-005, respectively.] (PsycINFO Database Record (c) 2007 APA, all rights reserved)

Note: Your library may have purchased access to this information through another service provider.